Brazil's 2026 election shows AI deepfakes slipping past election rules, warning Philippines

During Brazil's 2026 presidential election, Agência Lupa's VigIA project recorded 920 AI-generated posts between mid-August and late September, half of which were deepfakes. The Superior Electoral Court has prohibited deepfakes since 2024 and requires AI content to carry disclosure labels, yet 60% of the examined posts had no such notices and many were posted by anonymous accounts. Only a minority of disclosures were added voluntarily, with platforms providing most of the tags.

The study found that synthetic media was used to promote candidates, attack rivals, and spread humor, complicating fact-checking efforts. President Luiz Inácio Lula da Silva appeared most often in the AI material, while Senator Flávio Bolsonaro was involved in a court-ruled exception for a private event. Observers say Brazil's experience underscores the need for the Philippines to strengthen enforcement of its own AI election rules before the 2028 polls.